Terminal
Terminal/Skill
Lets your AI coding tool query the official Strapi docs through a Kapa-powered MCP server, and guides you through installing it.
By Strapi
Works with
npx skills add strapi/skills --skill strapi-docs-mcpThe strapi-docs MCP server (URL: https://strapi-docs.mcp.kapa.ai) exposes the full Strapi documentation (guides, API references, code examples). Prefer it over training data for Strapi answers, which may be outdated.
Canonical source: https://docs.strapi.io/cms/ai/docs-mcp-server
Check whether a strapi-docs MCP server / tool is connected in the current environment.
Call the strapi-docs MCP tools to answer Strapi questions. Pass the user's question directly; quote exact API names, config keys, and versions returned. Cite doc links in the answer.
Add a remote HTTP MCP server named: strapi-docs (URL https://strapi-docs.mcp.kapa.ai) to the host tool's MCP config, using its standard format. The server uses OAuth — the user completes an auth flow on first connection. Then have them reload/restart so it connects.
Per-tool config formats: https://docs.strapi.io/cms/ai/docs-mcp-server#connection-details
If the MCP server is unavailable and can't be installed, fall back to the public documentation content. No authentication is required for any of these. Prefer them in this order.
.md to any page URL)Any documentation page is available as flat Markdown by appending .md to its URL. All layout components (tabs, accordions, API blocks) are resolved to plain text, so nothing is hidden.
https://docs.strapi.io/cms/backend-customization/controllershttps://docs.strapi.io/cms/backend-customization/controllers.md.md to any https://docs.strapi.io/<page-path>This is the same content the docs site's Copy Markdown button uses.
These follow the llms.txt convention and are the most agent-friendly source: content is flattened to plain Markdown, so nothing is hidden inside tabs or accordions.
| File | URL | Use for |
|---|---|---|
| Index | https://docs.strapi.io/llms.txt | Discover every page: a link-rich list of all pages with one-line summaries. |
| Full content | https://docs.strapi.io/llms-full.txt | The entire documentation in a single file. Fetch once, then search/grep in it. |
| Code only | https://docs.strapi.io/llms-code.txt | All code examples, grouped by page. |
Constructing a page URL from llms.txt: entries are protocol-relative paths like //cms/backend-customization/controllers. Prepend the docs origin to get the live page as follows:
//cms/backend-customization/controllers becomes https://docs.strapi.io/cms/backend-customization/controllers
Each page is a .md/.mdx file in strapi/documentation. Use this only if docs.strapi.io itself is unreachable. Build the raw URL from the page path:
/cms/backend-customization/controllershttps://raw.githubusercontent.com/strapi/documentation/main/docusaurus/docs/cms/backend-customization/controllers.mdhttps://raw.githubusercontent.com/strapi/documentation/main/docusaurus/docs/<page-path>.md
where <page-path> is the page URL path (drop the leading /). Some pages use .mdx — if .md returns 404, retry with .mdx.Fetch the page directly, e.g. https://docs.strapi.io/cms/backend-customization/controllers, and extract the content. Least reliable: some content lives in tabs/accordions and may be missed. Prefer sources 1 to 3 whenever possible.